Pyroglutamylated amyloid-β peptide reverses cross β-sheets by a prion-like mechanism.

The amyloid hypothesis causatively relates the fibrillar deposits of amyloid β peptide (Aβ) to Alzheimer's disease (AD). More recent data, however, identify the soluble oligomers as the major cytotoxic entities. Pyroglutamylated Aβ (pE-Aβ) is present in AD brains and exerts augmented neurotoxicity, which is believed to result from its higher β-sheet propensity and faster fibrillization. While this concept is based on a set of experimental results, others have reported similar β-sheet contents in unmodified and pyroglutamylated Aβ, and slower aggregation of pE-Aβ as compared to unmodified Aβ, leaving the issue unresolved. Here, we assess the structural differences between Aβ and pE-Aβ peptides that may underlie their distinct cytotoxicities. Transmission electron microscopy identifies a larger number of prefibrillar aggregates of pE-Aβ at early stages of aggregation and suggests that pE-Aβ affects the fibrillogenesis even at low molar fractions. Circular dichroism and FTIR data indicate that while the unmodified Aβ readily forms β-sheet fibrils in aqueous media, pE-Aβ displays increased α-helical and decreased β-sheet propensity. Moreover, isotope-edited FTIR spectroscopy shows that pE-Aβ reverses β-sheet formation and hence fibrillogenesis of the unmodified Aβ peptide via a prion-like mechanism. These data provide a novel structural mechanism for pE-Aβ hypertoxicity; pE-Aβ undergoes faster formation of prefibrillar aggregates due to its increased hydrophobicity, thus shifting the initial stages of fibrillogenesis toward smaller, hypertoxic oligomers of partial α-helical structure.

Introduction

Alzheimer’s disease (AD) is a neurodegenerative disorder characterized by neuronal and synaptic loss leading to cognitive and memory impairment. Extracellular fibrillar deposits (plaques) of amyloid-β (Aβ) peptide have been found in the AD brain and thought to be causatively related to the disease.[1−3] However, currently accumulated evidence identifies the soluble oligomers of Aβ as the main neurotoxic entities.[4−8]

Aβ is a proteolytic product of the amyloid precursor protein and can contain varying numbers of amino acid residues, with the 40- and 42-residue peptides (Aβ1–40 and Aβ1–42) being the prevalent forms. Circular dichroism (CD) and NMR data indicate that in organic solvents such as hexafluoroisopropanol (HFIP) Aβ1–42 adopts a partially α-helical structure and in the presence of >80% H2O acquires a β-sheet structure.[9] In aqueous media, Aβ forms fibrils composed of β-sheets where the strand axis is approximately perpendicular and the H-bonding is parallel to the long fibrillar axis, known as a cross β-sheet structure.[10,11] Antiparallel β-sheets were proposed to constitute the core structural motif of fibrils formed by Aβ1–42 or its fragments.[12,13] However, solid state NMR studies on Aβ1–42 and shorter peptides identified in-register parallel β-sheet structures,[14−16] consistent with models derived from spin-label EPR,[17] solution NMR,[18] and Fourier transform infrared (FTIR) studies.[19,20] Apparent inconsistencies might originate from different stages of peptide aggregation in different samples, as Aβ1–42 oligomers and fibrils were shown by FTIR to adopt antiparallel and parallel β-sheet structures, respectively.[21]

Significant amounts of N-terminally truncated and pyroglutamylated (at Glu3 or Glu11) Aβ peptide (pE-Aβ) have been identified in AD brains and shown to aggregate at increased rates[22−26] and to be more cytotoxic than unmodified Aβ.[27,28] Even at low fractions, pE-Aβ coaggregates with Aβ by a seeding mechanism and forms structurally distinct and highly toxic oligomers.[27] Dot-blot experiments using conformation-sensitive antibodies showed that the highly toxic oligomers containing 5% or less pE-Aβ were structurally different from the mildly toxic unmodified Aβ aggregates of similar size.[27] While these studies imply a structural mechanism for augmented toxicity of pE-Aβ, the underlying structural differences between Aβ and pE-Aβ remain uncharacterized. Solution NMR showed that AβpE3–40 in trifuoroethanol/water (2:3) has a reduced α-helical propensity compared to Aβ1–40,[29] consistent with a significantly higher β-sheet content and faster fibrillogenesis of pE-Aβ.[23,25] Conversely, CD studies identified similar content of β-sheet in both unmodified Aβ and pE-Aβ peptides,[30] and fibrillogenesis of AβpE3-42 was reported to be significantly slower compared to Aβ1–42.[31] These conflicting data on fibrillogenesis of unmodified Aβ and pE-Aβ are evidently related to the inherent polymorphism and sensitivity of the Aβ peptides to the experimental conditions and procedures.[32,33] Since the content of pE-Aβ in AD brains varies in a wide range and affects the structure and the toxicity of the amyloid aggregates,[27,34,35] individual structures of unmodified Aβ and pE-Aβ separately and in combination, as well as the mutual structural effects, should be determined to shed light on the molecular mechanism underlying the altered fibrillogenesis of pE-Aβ.

Here, we have employed transmission electron microscopy (TEM), CD, and FTIR spectroscopy to analyze structural transitions in Aβ1–42 and AβpE3-42 peptides during fibrillogenesis. Isotope-edited FTIR was used to examine structural changes in both peptides combined in one sample, which allowed identification of the profound prion-like conformational effect of pE-Aβ on the unmodified Aβ. Specifically, pE-Aβ not only exhibited an increased α-helical and reduced β-sheet propensity but also was able to retard β-sheet formation by Aβ and to reverse β-sheets to α-helical structure at initial stages of fibrillogenesis. These findings suggest that the augmented cytotoxicity of pE-Aβ may result from its preference to form hypertoxic aggregates of partial α-helical structure as opposed to mildly toxic β-sheet fibrils.

Materials and Methods

Materials

The Aβ1–42 and uniformly 13C-labeled Aβ1–42 peptides were purchased from rPeptide (Bogart, GA, USA) and were >97% pure. AβpE3-42 was from Innovagen (Lund, Sweden) and was 98% pure. The peptides were analyzed by MALDI-TOF mass-spectrometry at the ICBR Proteomics Core Facility of the University of Florida (Gainesville, FL, USA), and the amino acid compositions of all three peptides were confirmed. Salts, buffers, HFIP, and other chemicals were from Fisher Scientific (Hanover Park, IL, USA) or Sigma-Aldrich (St. Louis, MO, USA).

Experimental Procedures

In all experiments, the lyophilized peptides were initially dissolved in HFIP at 200 μM concentration to disperse any preformed aggregates. In TEM experiments, appropriate amounts of the peptides were dried in a glass vial by desiccation for 15 min, followed by incubation in an aqueous buffer of 50 mM NaCl + 50 mM Na,K-phosphate (pH 7.2) at 37 °C with constant stirring for 24 h. TEM samples were prepared following the procedures described by Nilsson,[36] i.e., by deposition of 5 μL of peptide suspension on the grid, incubation for 5 min, and rinsing with 4 μL of distilled/deionized water, followed by staining for 30 s with 2 μL of 3% uranyl acetate, washing twice with 5 μL of distilled/deionized water, and air-drying. Grids for the negative control experiments were prepared by identical procedures using 5 μL of blank buffer instead of the peptide suspension. Images were acquired on a JEOL TEM-1011 operated at 80 kV using thin (<3 nm) holey carbon grids (Ted Pella, Inc., Redding, CA, USA).

In CD experiments, the HFIP solutions of peptides were dried by desiccation in a 4 mm × 4 mm quartz cuvette and spectra were collected between 180 and 330 nm to determine the structure of the dry peptides. Subsequently, an aqueous buffer of 50 mM NaCl + 50 mM Na,K-phosphate (pH 7.2) was added to a 50 μM final concentration of the peptides and spectra were acquired consecutively for 24 h to identify secondary structural changes upon fibrillogenesis at 37 °C with constant stirring, using a J-810 spectropolarimeter (Jasco, Tokyo, Japan). To improve the signal-to-noise ratio, the spectra were smoothed using a 13-point Savitzky-Golay linear least-squares algorithm embedded in the Igor Pro 5.03 software.

FTIR experiments were conducted to determine the structure of the peptides in desiccated form, nominally hydrated by atmospheric humidity, and in the presence of excess aqueous buffer. Desired amounts of the peptides were dissolved in HFIP at 200 μM concentration, and 40 μL of the solution was placed on a CaF2 FTIR window and dried in a desiccator for 15 min. FTIR spectra of the peptide samples were collected while the peptide was allowed to absorb humidity from the atmosphere as monitored by the increase in the H2O stretching band intensity around 3270 cm–1. Then, 80 μL of aqueous buffer (10 mM Na,K-phosphate in D2O, pD 7.2, corresponding to the pH-meter reading of 6.8) was added to the peptide and the sample was sealed by a second window using a 50 μm-thick Teflon spacer, followed by measurements of spectra of the peptide in aqueous medium. The spectra were measured by coadding 500 scans on a Vector-22 FTIR spectrometer (Bruker Optics, Billerica, MA, USA) equipped with a liquid nitrogen-cooled Hg–Cd–Te detector, at 2 cm–1 nominal resolution at 25 °C, as described earlier.[37] Reference transmission spectra were collected using either a single CaF2 window or the buffer sealed between two windows and were used to calculate the absorbance spectra. H2O vapor spectra were measured separately and subtracted from the sample spectra when necessary. The spectra were smoothed as described above, and baseline correction was applied.

Results

Based on earlier findings that AβpE3-42 forms aggregates that are structurally different from the aggregates of Aβ1–42 and exert prion-like toxicity on cultured neurons,[27] we hypothesized that AβpE3-42 modulates the structure of the unmodified peptide reminiscent of prions. Since the content of pyroglutamylated Aβ can vary up to 50% of total Aβ,[27,34,35] we studied pE-Aβ/Aβ samples at 1:9 and 1:1 molar ratios in addition to pure Aβ and pE-Aβ peptides. TEM images were acquired at 2, 4, 12, and 24 h of incubation, as described in the Materials and Methods. Most significant differences between Aβ1–42 and AβpE3-42 were detected at the early stages of aggregation. At 2 h, the samples of AβpE3-42 were dominated by nonfibrillar aggregates of irregular shape and average dimension of 30–100 nm, while the Aβ1–42 samples showed well-defined fibrils and a smaller number of small aggregates (Figure 1a and b). The 1:9 and 1:1 molar combinations contained predominantly prefibrillar structures (Figure 1c and d). At 4 h of incubation, the fibrils were seen in all samples, with little morphological differences, with small aggregates still present (Figure 1e–h). With progression of fibrillogenesis through 24 h, the small aggregates were converted to fibrils which became more extended and entangled (Figure 1i–p). While the mature fibrils formed by the AβpE3-42 peptide seem to be thicker, possibly bundled (cf. part n of Figure 1 with parts m, o, and p), consistent with earlier observations,[24,28] the TEM data do not allow identification of more distinct, definitive morphological differences between the fibrils of the unmodified and pyroglutamylated peptides and their combinations. Taking into account the clear differences between the early stage assemblies of Aβ1–42 and AβpE3-42, these data suggest that the fibrillogenesis of the two forms may follow different pathways, leading to fibrils that are similar at the level of morphology. The images obtained in negative control experiments showed clear grids, as expected (not shown).

Figure 1

TEM images of Aβ1–42 (a, e, i, m), AβpE3-42 (b, f, j, n), AβpE3-42/Aβ1–42 = 1:9 (c, g, k, o), and AβpE3-42/Aβ1–42 = 1:1 (d, h, l, p) incubated in aqueous buffer of 50 mM NaCl + 50 mM Na,K-phosphate (pH 7.2) for 2 h (a–d), 4 h (e–h), 12 h (i–l), and 24 h (m–p) at 37 °C with constant stirring. The horizontal bar in each panel equals 100 nm.

Earlier TEM studies showed similar morphologies of AβpE3-42 and Aβ1–42 aggregates at the initial stages of aggregation but more “curvilinear and entangled” fibrils of Aβ1–42 at 1–2 days of fibrillogenesis.[38,39] In the equimolar sample, the fibrils were less entangled, i.e. more like AβpE3-42 fibrils, suggesting that pE-Aβ might be able to dictate its morphological (and probably structural) features to the aggregates.

It has been recognized that the fibrillar morphology is determined by the molecular structure of the peptides,[23,30,38,39] but CD studies provided conflicting data on the relative secondary structural changes in Aβ and pE-Aβ during fibrillogenesis (see above). To monitor the structural transitions in the peptides during fibrillogenesis, peptide samples dried from HFIP solution were used as a starting point, before the onset of aggregation. CD spectra of Aβ1–42, AβpE3-42, and their combinations in dry form shown in Figure 2a indicate mostly α-helical structure with two minima around 222 and 208 nm.[40,41] The spectrum of AβpE3-42 has a significantly reduced ratio of ellipticities θ208/θ222, indicative of a more flexible or disordered α-helix.[42] These results concur with solution NMR data showing α-helical conformation for both Aβ1–42 and AβpE3–40 in organic solvents.[9,29] Upon addition of an aqueous buffer and incubation at 37 °C with constant stirring, the peptides undergo significant structural changes. Aβ1–42 promptly adopts and maintains β-sheet structure, as evidenced by a deep minimum at 215–216 nm of spectra measured at 1 and 24 h of incubation (Figure 2b). The spectra of AβpE3-42, on the other hand, show a wide well between 208 and 222 nm, most likely indicating a combination of α-helical and β-sheet structures (Figure 2c). These data suggest substantially different structures of Aβ1–42 and AβpE3-42, while the former readily adopts β-sheets, the latter shows increased α-helical propensity. The CD spectra of the 1:9 AβpE3-42/Aβ1–42 combination display β-sheet features, i.e. a prominent minimum at 216 nm at 1 h and at 219 nm at 24 h of incubation (Figure 2d). The higher intensity and the red shift of the signal at 24 h may reflect gradual suspension of the peptide into the aqueous medium and decreased solvent accessibility upon aggregation.[40] It should be noted that the spectra of Figure 2d are dominated by the structural features of Aβ1–42 which are present at a large molar excess (90%). At 1:1 molar ratio, the 1-h spectrum shows a minimum at 209 and a shoulder at 223 nm (Figure 2e), implying α-helix structure, possibly including a β-sheet component, as in the case of pure AβpE3-42 (cf. blue spectra in Figure 2c and e). At 24 h, the spectrum has a β-sheet minimum at 216 nm and a shoulder at 227 nm, likely generated by a turn structure. It is remarkable that AβpE3-42 exerts a dominant structural effect, especially at the early stages of fibrillogenesis. Thus, consistent with the TEM data, CD results indicate that (a) Aβ1–42 and AβpE3-42 evidently follow distinct structural pathways of fibrillogenesis and (b) AβpE3-42 is able to divert the overall path toward less β-sheet and more α-helical intermediates.

Figure 2

CD spectra of dry and water-suspended peptides. (a) Aβ1-42 (green), AβpE3-42 (turquoise), AβpE3-42/Aβ1–42 = 1:9 (blue), and AβpE3-42/Aβ1–42 = 1:1 (red) were dissolved in HFIP, followed by 15 min of desiccation in a 4 mm × 4 mm quartz cuvette and collection of the spectra. (b–e) Aqueous buffer of 50 mM NaCl + 50 mM Na,K-phosphate (pH 7.2) was added to a 50 μM final concentration of Aβ1–42 (b), AβpE3-42 (c), AβpE3-42/Aβ1–42 = 1:9 (d), and AβpE3-42/Aβ1–42 = 1:1 (e), and spectra were acquired after 1 h (blue) and 24 h (red) of incubation at 37 °C with constant stirring.

While TEM and CD data indicate distinct structural differences between Aβ1–42 and AβpE3-42 and suggest a dominant structural effect of pE-Aβ on Aβ, neither of these methods has the capability of resolving the individual structures of the two peptides in combination and the mutual structural effects. Individual structures of two proteins combined in one sample can be determined by FTIR spectroscopy if their amide I bands are spectrally separated, which is achieved by 13C-labeling of one of the proteins.[43−45] Despite its resolving power, such “isotope-edited” FTIR spectroscopy has not been used to characterize the concomitant structural transitions of Aβ and pE-Aβ during fibrillogenesis.

To detect the structural changes accompanying formation of amyloid fibrils, FTIR spectra were measured before and after exposure of the peptides to an aqueous buffer. Both peptides were dissolved in HFIP and dried on a CaF2 window by 15 min desiccation. AβpE3-42 adopts an intramolecular antiparallel β-sheet structure (peak at 1634 cm–1 and shoulder around 1695 cm–1), as well as a significant fraction of α-helix and turn structures (broad component(s) between 1685 and 1650 cm–1) (Figure 3a).[45−47]13C-Aβ1–42 forms an intermolecular β-sheet (main peak at 1588 cm–1) plus turns and an insignificant α-helix (component at 1617 cm–1) (Figure 3a). These data imply that 13C-Aβ1–42 readily forms a cross-β structure even in the absence of an aqueous medium while AβpE3-42 forms intramolecular β-hairpins and an α-helix. In aqueous (D2O) buffer, both peptides adopt parallel intermolecular β-structure, as evidenced by the major amide I peaks at 1628 cm–1 for AβpE3-42 and 1585 cm–1 for 13C-Aβ1–42 (Figure 3b). However, the prominent component between 1680 and 1650 cm–1 in the spectra of AβpE3-42 indicates that the pyroglutamylated peptide retains significant fractions of α-helical and turn structures. (The small peak in the spectra of 13C-A1–42 at 1673 cm–1 is likely generated by trace amounts of trifluoroacetic acid usually present in synthetic peptide samples.)

Figure 3

(a) FTIR spectra of AβpE3-42 (solid) and uniformly 13C-labeled Aβ1–42 (dotted) dried from a 200 μM HFIP solution on a CaF2 window. (b) FTIR spectra of the two peptides, as indicated, in 10 mM Na,K-phosphate in D2O, pD 7.2. Decreasing line darkness corresponds to time of exposure of the peptides to the buffer for 10, 30, 50, 70, 90, and 120 min.

The amide II spectral region provides additional structural information on proteins and peptides. Flexible secondary structures or open, solvent accessible tertiary structures undergo faster amide hydrogen/deuterium exchange resulting in reduction of the amide II band intensity around 1540 cm–1.[45−47] A considerable amide II band is retained in the spectrum of 13C-Aβ1–42 after a 2 h exposure to D2O while that of AβpE3-42 is lost (Figure 3b), indicating Aβ1–42 forms a rigid secondary structure and/or a tight, solvent-inaccessible tertiary structure, characteristic of a cross β-sheet structure,[10,11] while AβpE3-42 has a more open tertiary structure and/or more flexible secondary structure.

It has been shown earlier that isotope-edited FTIR can be used to probe the intermolecular interactions of peptides.[43,44] In the case of closely spaced 13C-labeled peptide units, through H-bonding or through space 13C–13C vibrational coupling between adjacent strands results in a lower frequency (∼1590–1594 cm–1) amide I mode whereas 13C–12C coupling between labeled and unlabeled units generates higher frequency (∼1600–1604 cm–1) components of enhanced intensity.[20,48,49] FTIR studies were conducted on combined 13C-Aβ1–42 and unlabeled AβpE3-42 to probe (a) the intermolecular interactions and (b) mutual structural effects of the peptides. Since pE-Aβ in AD brain can constitute up to 50% of total Aβ,[34,35] we studied AβpE3-42/13C-Aβ1–42 samples at 10% and 50% molar fractions of AβpE3-42. Data of Figure 4 indicate that the β-sheet peak of 13C-Aβ1–42 at 1585 cm–1 up-shifts by 3 and 10 cm–1 in the presence of 10% and 50% pE-Aβ, respectively, while the β-sheet peak of AβpE3-42 at 1626–1628 cm–1 up-shifts by 10 and 4 cm–1 in the presence of 90% and 50% 13C-Aβ1–42, respectively, indicating strong interactions and vibrational couplings between the two peptides. Thus, AβpE3-42 and Aβ1–42 form a mixed β-sheet structure with tight intermolecular interactions.

Figure 4

FTIR spectra of AβpE3-42 and 13C-Aβ1–42 combined at 1:9 (a) and 1:1 (b) molar ratios, incubated in a D2O-based phosphate buffer (pD 7.2) for 2 h, at a total peptide concentration of 100 μM. Black and gray lines are the experimental spectra obtained on the two peptides combined in one sample and the weighted sums of individual spectra, respectively. The weighted sums were obtained as A = ∑fA, where f is the molar fraction and A is the absorbance spectrum of each individual peptide measured separately.

Next, we tested the emerging hypothesis that the pyroglutamylated peptide is able to modulate the structure of the unmodified Aβ during amyloid fibril formation. To assess early structural events in aggregation, combinations of HFIP solutions of AβpE3-42 and 13C-Aβ1–42 were dried on a FTIR CaF2 window followed by collection of spectra while the sample was allowed to absorb atmospheric humidity. Figure 5a shows the spectra of AβpE3-42 and 13C-Aβ1–42 combined at 1:9 molar ratio in a wide spectral range. The 1700–1500 cm–1 region corresponds to the amide I and amide II modes and reflects the peptides’ secondary and dynamic structure whereas the signal in the 3450–3150 cm–1 region results from the amide A and H2O stretching modes.[45−47] Spectra shown in gray and black solid lines in Figure 5a and b were collected on a sample that was dried by desiccation and exposed to the atmosphere for 10 and 20 min, respectively. The peptides absorb atmospheric humidity, which results in an increase in the H2O stretching band around 3270 cm–1. Nominal hydration causes conformational transitions in both peptides and corresponding spectral changes in the amide I region, shown in Figure 5b. At a 1:9 AβpE3-42/13C-Aβ1–42 molar ratio, the amide I band is dominated by the spectral features of the latter peptide, i.e. α-helical and β-sheet components at 1617 and 1592 cm–1, respectively (Figure 5b). The broad band around 1655 cm–1 evidently results from the overlapped α-helical mode of AβpE3-42 and the turn structures of 13C-Aβ1–42. As the sample absorbs atmospheric moisture, 13C-Aβ1–42 undergoes α-to-β transition (intensity transfer from 1617 to 1592 cm–1, cf. gray and black solid lines in Figure 5b). However, a more humid 1:9 mixture (black solid line) contains less β-sheet and more α-helix than expected without interaction between the two peptides (the weighted sum of individual spectra shown in a dotted line). These and above data suggest that pE-Aβ slows down cross β-sheet formation in Aβ by direct intermolecular interactions.

Figure 5

FTIR spectra of AβpE3-42 and uniformly 13C-labeled Aβ1–42 at 1:9 molar ratio (a and b) and 1:1 molar ratio (c). Gray and black solid lines are experimental spectra of a sample prepared in HFIP, followed by solvent removal by desiccation and exposure to atmosphere for 10 and 20 min, respectively. The dotted spectrum is the weighted sum of the spectra of each peptide measured individually, exposed to the atmosphere for 15 min. Construction of the weighted sum spectra is described under Figure 4. Panel b is a zoom-in into the amide I/II region of spectra shown in panel a.

Moreover, at 1:1 molar combination, pE-Aβ reverses the cross-β structure and hence fibrillization of Aβ. As shown in Figure 5c, in the presence of 50% pE-Aβ, the intermolecular β-sheet peak of 13C-Aβ1–42 shifts from 1588 to 1595 cm–1 (cf. dotted and gray spectra), indicating strong interaction between the two peptides. (In the presence of only 10% AβpE-42, a smaller shift from 1588 to 1592 cm–1 is observed, Figure 5b.) In the presence of an equimolar amount of AβpE3-42, there is no α-helix to β-sheet conversion of 13C-Aβ1–42 over time, as seen by the similar signal intensity at 1617 cm–1 in gray and black solid spectra in Figure 5c. Most importantly, AβpE3-42 causes a strong reduction of the intermolecular β-sheet signal of 13C-Aβ1–42 at 1595 cm–1 during longer coincubation (cf. gray and black solid spectra in Figure 5c) paralleled with increased intensity around 1658 cm–1. The spectra of the combination (gray and black solid lines in Figure 5c) indicate an increase in α-helical structure (signals at 1658 and 1617 cm–1) and a decrease in β-sheet structure (signals at 1634 and 1585–1588 cm–1) in both peptides as compared to the weighted sum of individual spectra (dotted line in Figure 5c). Although the component at 1658 cm–1 might partially result from turn structures in addition to the α-helix in AβpE3-42, these data identify prominent mutual conformational effects of the two peptides; pE-Aβ at 10% delays cross β-sheet formation and hence fibrillization and at 50% reverses the cross β-sheet structure formation of Aβ.

Discussion

Our data identify significant differences between AβpE3-42 and Aβ1–42 at the levels of morphology as well as secondary and tertiary structures. At the initial stages of fibrillogenesis, the pyroglutamylated AβpE3-42 peptide forms more prefibrillar aggregates, apparently due to its increased hydrophobicity, and it undergoes fibril elongation slower than the unmodified Aβ1–42 peptide (Figure 1a,b), in agreement with earlier data.[31] Retardation of fibrillization appears to be imparted to Aβ1–42 even at low molar contents of AβpE3-42 (Figure 1c). CD data indicate augmented α-helical and diminished β-sheet propensity of AβpE3-42, which is transmitted to the mixed assemblies (Figure 2). FTIR indicates that the unmodified peptide readily forms a tightly packed intermolecular β-sheet, while pE-Aβ forms a less compact β-structure and contains more α-helix and turn structures than Aβ (Figure 3). Furthermore, the pyroglutamylated peptide not only exhibits a significantly lower tendency to form a β-sheet structure but also inhibits cross β-sheet formation in the unmodified peptide through direct interactions (Figures 4 and 5). These structural transitions occur rapidly upon hydration, but they can be captured when the peptides undergo nominal hydration by exposure to atmospheric humidity. These conditions are both technically beneficial and meaningful because the fibrils formed by Aβ1–40 were shown to contain an extremely low fraction of water, i.e., an average of 1.2 water molecules per β-strand.[50] Significant retention of the amide II band in the spectrum of Aβ1–42 in a D2O-based buffer (Figure 3b) is in line with this finding. Furthermore, the rapid loss of the amide II band of AβpE3-42 upon exposure to D2O indicates a more flexible secondary and/or a more open tertiary structure of the pyroglutamylated peptide.

If pE-Aβ eventually forms fibrils that contain β-sheet structure, even though different from the fibrils formed by unmodified Aβ, why do the intermediate structural steps matter? The answer is that the oligomeric, prefibrillar assemblies of Aβ that adopt still poorly characterized “pathological conformation“ are the most toxic species.[51] For example, the secreted pool of Aβ oligomers exerts its toxic effect partly by binding to a set of receptors, including the insulin receptor that recognizes an α-helical ligand.[51] Intracellular oligomers bind to the mitochondrial or endoplasmic reticulum proteins and cause cell damage through oxidative stress or calcium dysregulation before they are secreted.[52,53] Since the cytotoxic effect is exerted before formation of the extracellular deposits, chracterizaton of the intermediate “pathological conformations“ is crucial.

The structural impact of pE-Aβ on the unmodified peptide even at low molar content (10%) is indicative of a prion-like effect. The pE-Aβ peptide tightly interacts with Aβ, as indicated by efficient 12C–13C vibrational coupling (Figures 4 and 5), and thereby transmits the specific structural features to the unmodified peptide. Data of Figures 2–5 strongly imply that this specific structure is rich in α-helix as opposed to β-sheet. This effect takes place even at 10% pE-Aβ, when pE-Aβ molecules cannot simultaneously interact with a large excess of unmodified Aβ. We therefore propose that once the structural transition occurs in the unmodified Aβ molecule by its interaction with pE-Aβ, it acquires the capability to further transmit the altered structure to other Aβ peptides by direct interaction. This prion-like conformational effect of pE-Aβ may eventually shift the overall path of peptide aggregation toward formation of hypertoxic, lower molecular weight aggregates of partial α-helical structure and thus suppress formation of less toxic cross β-sheet fibrils.